[Windows] Compare the loudness of two EQed signals

[Jose Hidalgo]

Hi everybody,

I have two different Equalizer APO presets and I'd like to know their relative loudness for volume-matching purposes.
For example by playing some pink noise and switching EAPO presets.
Since I'd like some precise measurements as a clean basis, I won't be trusting my ears on this one (plus pink noise sounds terrible anyway).
There are loudness meter apps such as this one ( https://youlean.co/youlean-loudness-meter/ ), but I wonder how I could put it in the signal path AFTER Equalizer APO, so it analyzes the EQed signal, not the raw one. Am I missing something ?

Thanks in advance for your help.

 

[Jose Hidalgo]

... hello ?... *echoing voice in the darkness*

 

[Jose Hidalgo]

I guess I'll have to keep bumping this topic until someone finally replies. *fingers crossed*

 

[RayDunzl]

An interesting question...

With, I'm sure, more than one answer, that will not necessarily lead to "equal loudness" you seek.

---

If comparing EQ, I might pick 1kHz and adjust for equal level there.

That will give you the differences if they exist at all other frequencies.

Not what you want?

Oh well.

 

[dougi]

Try the free Orban meter software. It has a variety of loudness based meters. It should be able to be set to monitor the Windows output.

 

[pozz]

You can probably use this: https://vb-audio.com/Cable/

Routing tool for Windows.

 

[Jose Hidalgo]

@dougi , already tried. Look at the reply I was writing to RayDunzi.

-----

There are already standards, as you certainly know, so we don't need to reinvent the wheel.
More specifically, I was implying something like LUFS.
There are many loudness meters that use the LUFS standard, including free ones. For example last week I have installed Orban Loudness Meter.

My real problem is that I need to be able to analyze the EQed signal, so after Equalizer APO.
Currently I don't seem to be able to do that. Orban Loudness Meter seems to analyze the signal before, because when I switch EQs I see no difference.

So that would be my real question (how to achieve that, and with what software).

Not what you thought my question would be ?

Oh well.

 

[Jose Hidalgo]

@pozz , I can use VB-Cable indeed. Well, in my system I have uninstalled it because of unexplainable audio glitches that we couldn't correct, despite a long discussion with Vincent Burel. But I could install Voicemeeter, it's pretty much the same thing. Hopefully it will work better.

More specifically, how would you do in order to be able to analyze the EQed signal ? It's more a Windows question than an audio question.

I should add that the developer of Equalizer APO also had a similar thing in mind 2 or 3 years ago. He wanted to improve the already existing VST host system within EAPO, so that apps like loudness meters in VST form could work with EAPO. AFAIK currently it's still not possible.

 

[RayDunzl]

I put a track into Audacity, EQ'd it, compared the RMS level to the original, amplified the EQ'd version, using the difference in RMS, and the resulting RMS matched between the original and the modified.

Are they the same loudness?

I don't know.

 

[Jose Hidalgo]

Probably because RMS and LUFS aren't the same thing.
If I wanted something like RMS, I would have done like you did : Audacity, VUmeters, etc.
Hence the title of this topic : loudness (as in "perceived loudness according to the LUFS standard"), and not RMS.

For future readers, here are the differences : https://integraudio.com/difference-between-lufs-rms-true-peak-loudness-meters-explained/

 

[pozz]

I don't know the Windows audio stack that well.

Here's how I would do it: apply the PEQs to a signal of your choice in a DAW in multiple tracks. Have the loudness meter in a VST and assign an instance to each track, press play and watch the result. Not necessarily pink noise either. Anything, really. The difference as certain musical passages are run through would be interesting.

Personally I wouldn't use a LUFS meter for your purpose. It's designed for loudness normalization in commercial music releases, not meaningful comparisons of EQs and the like. But I don't think there is anything better that's very available. You'd have to get deep into signal processing and the science direct and generate complex loudness representations.

At least this is my current thinking.

 

[dougi]

@dougi , already tried. Look at the reply I was writing to RayDunzi.

-----

There are already standards, as you certainly know, so we don't need to reinvent the wheel.
More specifically, I was implying something like LUFS.
There are many loudness meters that use the LUFS standard, including free ones. For example last week I have installed Orban Loudness Meter.

My real problem is that I need to be able to analyze the EQed signal, so after Equalizer APO.
Currently I don't seem to be able to do that. Orban Loudness Meter seems to analyze the signal before, because when I switch EQs I see no difference.

So that would be my real question (how to achieve that, and with what software).

Not what you thought my question would be ?

Oh well.

What about via loopback from an audio interface? I think the Orban meter can do the input as well?

 

[Blumlein 88]

In Toole's book on matching speakers which obviously don't have common frequency response, he suggests matching with pink noise. With a high pass 12/octave filter at 500 hz and a low pass 12/octave filter at 2000 hz.

 

[Jose Hidalgo]

I think I have found a workaround.

Instead of playing music (or pink noise) via my audio player and setting feedback loops, I have discovered that Orban accepts audio files too ("Analyze" tab).
So I just took some pink noise .wav, and with Audacity and my VST EQ I created several "EQed pink noise" files.
Then I just had to feed Orban with them and analyze the obtained LKFS values.
And by that I mean compare them, keeping in mind that all EQ presets have a negative preamp gain of course.

I'm doing it right now, and hopefully I will get some useful info.
Feel free to do it yourselves with your own EQ presets, it's a simple process and maybe we can exchange our impressions.

 

[thewas]

In the end 2 different EQ settings will always sound differently loud depending on the spectrum of the music played due to the different FR curves, so it will be never 100% transparent. Due to spectrum of most music I would personally do it with pink noise, but again this also just one of all other possible compromises.

 

[Thalis]

I use a very simple VU meter for my Win 7 set up and it responds to EQ changes. The meter itself may not be calibrated but at least I do see level changes. Not sure why tho and where it sits in the signal path.

 

[Jose Hidalgo]

Just wanted to report back.

Ultimately I didn't find the Orban results to be consistent with my subjective listening impressions. The loudness gap was just too important.
I tried with pink noise, and also with several real song files. I don't know if I did something wrong, but the results weren't good IMHO.

So in the end, with some time and patience, I did everything by ear, for each one of my headphones, by carefully listening to reference songs, switching EAPO presets instantly by keyboard shortcuts, and repeating the procedure many times. Below are my subjective findings :

• Audioquest Nighthawk (Wood or Carbon) : Oratory preset gain = -4.5 dB // "no EQ" gain for equal loudness = -8.5 dB (-4.0 dB vs. preset)
• Meze 99 Classics : Oratory preset gain = -7.2 dB // "no EQ" gain for equal loudness = -5.2 dB (+2 dB vs. preset)
• Hifiman Sundara : Oratory preset gain = -7.0 dB // "no EQ" gain for equal loudness = -5.5 dB (+1.5 dB vs. preset)
• Sennheiser HD600 : Oratory preset gain = -5.0 dB // "no EQ" gain for equal loudness = -5.5 dB (-0.5 dB vs. preset)

FYI the Oratory presets are those that aim for the Harman curve, not the Optimum Hifi one. I like bass too much to just let it go.

The listening level was average, not particularly loud. I'd say "comfy for a long listening session". For critical listening and/or short sessions I could push the volume up a little.

I was quite surprised by the Nighthawk results, but that's really what I hear. My guess is that the default Nighthawk sound has so much bass emphasis (up to 650 Hz, so that includes an important part of the midrange where a lot of vocals are) that it needs to be attenuated by -4.0 dB to get the same perceived loudness as the EQed sound. That being said, it was the most difficult measurement to make, because the EQed response is really different from the default one, so it's not easy to compare the two. I did this one many times, always ending up at -4 dB.

This is the default Nighthawk curve vs. the linearized Harman curve (green line), AFTER the -4.0 dB correction. The dotted curve is the EQed one.

For the 99 Classics it's completely different, because the default 99 Classics frequency response drops quite sharply between 200 and 400 Hz, from +5 dB to -5 dB (relatively to the Harman curve) and then stays between -5 dB. and -3 dB. So ultimately despite the default bass emphasis, the default midrange is so recessed vs. the Harman curve (more than the Nighthawk actually - I don't know why some people complain about the Nighthawk when the Meze default frequency response is way worse ! ) that the whole default sound needs to be pushed up 2 dB to match the perceived loudness of the EQed sound.

This is the default 99 Classics curve vs. the linearized Harman curve (green line), AFTER the +2.0 dB correction. The dotted curve is the EQed one.

So in the end I think my results are about consistent. If anybody manages to do better, I'd love to try his findings.

 

[dasdoing]

if you put a VST last in the chain in EQAPO wont it show the processed output?

 

[dasdoing]

the Orban one actualy analyses the output of EQAPO